1 Remember from the presentation on Fundamentals of ASR we learned - - PDF document

1

Remember from the presentation on “Fundamentals of ASR” we learned that there are three requirements for ASR expansion to occur; these are: A sufficient quantity of reactive silica … which is provided by the aggregate A sufficient quantity of alkali … which is supplied … predominantly … by the portland cement And a supply of moisture during service If li it f th i t h ld b bl t t d i If we can limit one of these requirements … we should be able to prevent damaging ASR from happening

2

For a consideration of these three requirements we can produce a list of measures that might work Use of a non-reactive aggregate ensures that there is insufficient reactive silica available in the concrete We should really consider the second and third bullet together … Limiting the alkali content of the concrete – perhaps through the use of low alkali cement – ensures that there is insufficient alkali for damaging ASR The use of supplementary cementing materials can also be seen as a means for limiting the alkali content … in a way … although the use SCMs may not reduce the alkali content of the concrete it can reduce the availability of the alkalis for reaction … this will be discussed later The fifth bullet concerns the use of chemical admixtures – namely lithium-based

• compounds. This doesn’t seem to be helpful in reducing any of our three

requirements for ASR … in fact lithium is an alkali. However, lithium works by changing the path of the reaction as we will see later. Note that limiting water is not on this list. Generally it is not practical to control the exposure condition of a civil-engineering structure such as a pavement or bridge. It might be our only option when dealing with an existing structure that already has ASR and therefore already contains a sufficient amount of alkali and reactive silica whi hich h can`t b be removed d – but wi h ith new constructi ion i it i is easi ier to ad dopt one of f th he measures listed here than to eliminate water.

3

The use of a non-reactive aggregate is, perhaps, the most obvious strategy. However … how sure are we that aggregate is really non-reactive. We have to put a great deal

• f faith in our test methods to tell us the correct answer. We looked at some of the

symptoms of ASR in this structure in a previous presentation. Well the aggregate used for this structure was tested by appropriate test methods of the day – and was deemed to be non-reactive. Sixty years later the dam is 7 inches taller than when it started!!!!

4

Of course – some of our more rapid tests fail a great many aggregates that are non- reactive and there may not be a suitable aggregate that passes this test in a given geographic area – so – in the absence of data from more reliable but longer-term tests – we have to assume the aggregates are reactive and adopt preventive measures if we want to use them. In some locations most of the aggregates may really be reactive and there may be no choice but to use them with appropriate preventive measures

5

Returning to the ubiquitous Mactaquac Dam … here is a structure that will be rebuilt in less than 20 years and the owners are considering using the same reactive aggregate once again. The reason for this is one of economy … rebuilding the concrete structures will require approximately half a million cubic yards of concrete or about three quarters

• f a million tons of aggregate. Sufficient aggregate will be produced by the

excavation of bedrock required for the new structure and – although this same rock i ibl f th i ti bl h t fi d t it i is responsible for the existing problems – we have to find a way to use it again.

6

Stanton developed the first expansion test for ASR. He produced mortar bars using different cements and various reactive aggregates and measured the expansion of these bars when exposed to moisture. He found that the expansion with a particular aggregate was strongly influenced by the alkali content of the cement. Typical expansion results after 2 years are shown for mortar bars made with a highly reactive sand from Ventura County, California. Excessive expansion and cracking of the mortar bars only occurred with cements with alkali contents in excess of 0.7 percent N-A-2-O-E. It was from this work that in 1940 – Stanton made the recommendation that damaging ASR was unlikely provided that the cement alkalis were below 0.6% sodium equivalent. This defined the classification of low-alkali cements in the United States – low- alkali cements being cements with less than or equal to 0.6% equivalent alkalis. Many jurisdictions in the U.S. still specify the use of low-alkali cement as means of preventing damaging ASR. As we will see later in the course, such a measure is NOT sufficient to guarantee that damage does not occur.

7

This slide shows the breakdown by equivalent alkali of 69 sources of Type one Portland cement from the United States – Canada – and Mexico. The equivalent alkalis can range anywhere from 0.1 percent to 1.2 percent with just under half the values being higher than 0.60% equivalent alkalis

8

There are a number of structures that have suffered from ASR despite the use of low-alkali cement

9

This is a well-known case of a pavement constructed with a cement with less than half a percent of alkalis

10

A review of the literature will show that this is not an isolated case

11

We now know that we have to control the alkali content of the concrete not JUST the alkali content of the cement … this was discussed earlier

12

There is a threshold alkali content below which expansion may not occur with a given aggregate – this threshold will vary from one aggregate to another. Currently there is no test method for determining the threshold alkali level for a given aggregate and – as discussed previously – aggregates will generally react and cause expansion and cracking at lower alkali contents in the field than are required in laboratory tests such as the concrete prism test

13

Remember this example that was used before – blocks expand in the field with less than four pounds per cubic yard of alkali – but concrete prisms from the same mix do not expand in the laboratory

14

Here are some more examples of ASR in a dam with less than 3 kg or 5 pounds of alkali

• another dam with less than 2 kg or just slightly more than 3 pounds
• And a pavement with less than 2 kg or 3 and a half pounds of alkali per cubic metre
• f concrete

I C In Canada – a range of alkali li li limits are used d d dependi di ng on th th t e type of f aggregat te and d f lk it the nature of the structure and its exposure condition

15

As a consequence of this … specifications such as the Canadian spec and the more recent AASHTO (ash-toe) recommended practice base alkali limits largely on field experience – in AASHTO the actual limit varies from 3 to 5 pounds depending on the risk of asr and the nature of the structure – this will be explained later

16

SI alternative to previous slide As a consequence of this … specifications such as the Canadian spec and the more recent AASHTO (ash-toe) recommended practice base alkali limits largely on field experience – in AASHTO the actual limit varies from 1.8 to 3 kilograms depending

• n the risk of asr and the nature of the structure – this will be explained later

17

Ok so what about s-c-m’s (ess-see-ems) Commonly used SCMs in North America include fly ash – ground granulated (iron) blast-furnace slag – silica fume and various natural pozzolans such as calcined clay

• r shale – such as metakaolin … and occasionally some volcanic ash and other

materials

18

Thomas Stanton recognized the possibilities of using pozzolans to control ASR in his seminal paper in 1940. In this and later work he showed that the impact of pozzolans went beyond the effect of merely diluting the cement alkalis in the mix.

19

[Not to instructor: animation used in the slide] SCMs include fly ash, slag, silica fume and natural pozzolans and almost all sources

• f these materials can be used to control ASR provided they are used in sufficient

quantity – so the question is – “How much is enough?” – to which the answer is … “It depends!” What does it depend on? Well of course it depends on – among other things – the composition of the SCM itself. Paradoxically – the main parameter that effects the efficiency of an SCM in terms of controlling ASR is the amount or reactive silica in the SCM. Those with a lot of silica – like silica fume – will behave like the left hand curve reducing expansion to a safe level at relatively low levels of replacement – such as 10 to 15% for silica

• fume. Those SCMs with lower amounts of silica and more calcium – such as Class

C fly ash or slag – have to be used at much higher replacement levels (maybe 50%

• r more) to control expansi

ion – lik like th he ri igh ht h hand d curve. The alkali content of the SCM is also important – those with higher alkali contents tend to be less effective. In addition to the composition of the SCM itself – the other parameters that affect the amount of SCM you need are the reactivity of the aggregate and the amount of alkali in the system – basically as the aggregate reactivity increases or as the amount of lk f alkali li i in th he system i increases – so too does the level l of f SCM requi ired d.

20

OK – lets look at some real data – this shows the expansion with time – up to 2 years – for concrete prisms produced with a reactive siliceous limestone – the Spratt aggregate. The control samples with no SCM expand rapidly exceeding the commonly-used expansion limit of zero-point-zero-four percent after a month or so. Partial replacement of 25% of the cement with a typical low-calcium Class F fly ash is extremely effective in reducing expansion even though the fly ash itself contains more alk lkali li th than the portl tland cement it t it repl laces. th d

21

The efficacy of low-calcium, Class F fly ash has also been well-established by exposure site studies … such as this one in the UK. These blocks are each a cubic yard of concrete – the control sample on the left exhibited deleterious expansion and cracking after about 5 years and was extensively cracked at 18 years. The block on the right has the same amount of portland cement plus 25% fly ash – no expansion

• r cracking was in evidence when it was inspected at 18 years.

22

The long-term beneficial effect of fly ash has also been established by field

• performance. This slide shows the Lower Notch Dam in Ontario – which was

complete around 1970

23

When it was built – it was known that the aggregate was reactive as it has caused deleterious expansion and cracking in a great many highway and hydraulic structures in the same location as the new dam

24

Expansion testing prior to construction showed that the combination of the local cement which was high in alkalis and fly ash performed as well as a low-alkali cement – since the low-alkali cement was from out-of-province – a decision was made to go ahead with the high-alkali cement plus fly ash combination. A recent inspection revealed no signs of ASR as the structure reached its fortieth birthday

25

[Animated slide] This slide shows two dams that were built as part of the same hydroelectric scheme around 1960 – the middle structure – an arch dam is extensively cracked and has required structural stabilization to prevent the arch moving upstream as it expanded – the structure upstream – a gravity dam – used the same aggregate and a cement of similar alkali content but also incorporated 25% fly ash to control heat – the fly ash has effectively staved off ASR expanion for more than 50 years now.

26

Not all fly ashes behave the same way. This slide shows that fly ashes with higher calcium contents do not completely suppress expansion when used at a replacement level of 25%

27

Indeed – the 2-year expansion of concrete prisms containing 25% fly ash is strongly dependent on the calcium content of the fly ash. Fly ashes with less than 20% calcium are generally effective unless they have a high alkali content. As the calcium content of the fly ash increases above 20% the efficacy of the fly ash decreases

28

High-calcium Class C fly ashes can be used … but have to be used at a much higher replacement rate as shown here.

29

This graph shows data from laboratory expansion tests using concrete prisms in accordance with ASTM C 12-93. The data show expansion curves for 5 concretes with varying levels of slag. The expansion decreases as the amount of slag in the mix increases. More than 35% slag is required to meet typical specification limits of zero-point- zero-four % at 2 years when combined with this highly reactive aggregate.

30

This graph shows test results for four different reactive aggregates – in this case the expansion at 2 years is plotted against the level of slag. As can be seen – when there is no slag in the mix – all of the 2-year expansion results are greater than 0.04% The expansion decreases with increasing slag content in all cases. Although 50% slag is needed to reduce the expansion to less than zero-point-zero- four percent in the concrete with the highly-reactive siliceous limestone – only 35% four percent in the concrete with the highly reactive siliceous limestone only 35% slag is required with the moderately reactive greywacke aggregate. Generally, the amount of slag required to control expansion increases as either the reactivity of the aggregate or the amount of alkali in the mix increases.

31

These data show that silica fume is very effective in controlling expansion due to ASR at relatively low levels compared with fly ash or slag. However, in the standard test conditions – the amount of silica fume required to control expansion with a highly-reactive aggregate is around 10% or more. This level of silica fume is higher than that typically used in concrete and can lead to problems with placing and finishing. The efficacy of silica fume appears to be strongly dependent on the alkali available in the concrete system and other studies have shown that lower amounts of silica fume can be used when the alkali content of the concrete is lower than typically used in these concrete prism tests

32

This is demonstrated in this slide. These are data for 12-inch concrete cores stored

• utdoors in South Africa. The graph on the left show expansion curves for concrete

with a lower alkali content – under 7 pounds – and it would seem that 7% silica fume is effective in controlling expansion – at least to 7 years. On the right are data for concrete with a greater alkali loading – above 8 pounds – and it can be seen that 7% silica fume was not effective in the long term.

33

34

Ternary blends containing two SCMs can also be effective measures for controlling

• ASR. Generally the effects of each SCM are additive. So you could use half the safe

level of one SCM combined with half the safe level of a second SCM – or a third of

• ne plus two thirds of the other – etcetera

35

So … how do scms work. Well to understand this we have to go to pore solution studies – the pore solution of concrete can be extracted by squeezing the concrete using very high pressures of 450 MPa or more – that is 65 thousand psi or more –

• nce extracted the solution can be analyzed to determine how much alkali is in

solution and therefore available for reaction with the aggregate.

36

– In this slide it can be seen that the pore solution extracted from a control cement paste is very rich in hydroxyl ions (remember these are the negatively charged ions that balance the positively charged sodium and potassium ions) – the concentration is almost most one mole per litre – which produces a pH of 14 – very caustic if you remember your pH scale. The incorporation of fly ash reduces the amount of alkali in solution – this means that even though fly ash itself contains alkalis – the use of fly ash reduces the amount of alkali that is AVAILABLE to the pore solution and ABLE to react with the aggregate In effect the fly ash – like other SCMs consumes ABLE to react with the aggregate. In effect the fly ash like other SCMs consumes

• alkalis. The higher calcium fly ashes are less effective in this role and are also less

effective in controlling ASR expansion.

37

Silica fume reduces the alkalinity of the pore solution very early on but there is some evidence that some of this alkali may be released later

38

We can prevent this later release by using silica fume together with fly ash (or slag) – relying on the faster reacting silica fume to consume the alkalis at early age and the fly ash to maintain a low alkalinity long term

39

• In essence what is happening is that the silica in the SCM is reacting with the alkalis

in much the same manner as the aggregate will – but because the SCM is so fine – it reacts faster and consumes the alkalis before they can react with the aggregate. Any reaction product that is formed is finely dispersed throughout the paste and large accumulations that may cause local swelling and expansion of the concrete do not

• ccur. We can achieve the same effect by grinding reactive aggregate very finely –

this is shown in the graph. The blue line shows the expansion of mortar bars produced with sand-sized Pyrex glass and high-alkali cement The red line shows produced with sand sized Pyrex glass and high alkali cement. The red line shows the same mix but with some of the Pyrex ground very finely and used as part of the

• cement. The ground Pyrex prevents reaction with the sand-sized Pyrex by

consuming the alkalis first – in fact, the ground Pyrex behaves like a pozzolan. In much the same way that ground silica beaves like a pozzolan – a coarse pozzolan can act like reactive aggregate. This is shown in the photograph – the reacting aggregate particle causing damaging is in fact an agglomeration of silica fume – literally millions of silica fume particles are stuck together to form one sand-size particle of reactive silica.

40

So how do we test to determine how much SCM we need? Well I think we already covered this in the previous presentation

41

But as a reminder – what we would like is to have aggregates that behave well in

• ur accelerated tests so that when we use the rapid 14-day test to determine the safe

level of SCM we get a similar result to that obtained by the longer-term concrete test in 2 years

42

In the AASHTO spec – it is strongly recommended that you calibrate the rapid mortar test with the concrete test for each aggregate – for this aggregates that give similar results in both tests – it is acceptable to use the rapid tests to evaluate the efficacy of preventive measures.

43

The potential for lithium-based chemicals to control expansion due to ASR dates back half-a-century to the work of McCoy and Caldwell at the Portland Cement Association in Chicago. They evaluated the effect of more than 100 chemicals on the expansion of mortars containing Pyres glass – which serves as a highly reactive artificial aggregate – they discovered a number of lithium compounds including – lithium carbonate, sulfate, chloride, silicate, fluoride and lithium nitrate – to be the most effective.

44

The graph shows some of the data from this early study. The expansion of mortar bars with lithium are plotted on the vertical (or y) axis relative to the expansion of the mortar bars without lithium – so a value of 1 on this axis means that the mortar expanded the same amount as the control – in other words there was no reduction. On the horizontal (or x-) axis of the graph the lithium content of the mortar is plotted as a ratio of the sum of sodium and potassium alkalis. In fact the number of moles of lithium is plotted as a ratio of the number of moles of sodium plus potassium. As this molar ratio increases – in other words as the content of lithium increases compared to the sum of the sodium and potassium – the relative expansion of the mortar bars decreases, McCoy and Caldwall’s data indicate that a molar ratio of 0.74 might be sufficient to suppress expansion to a suitable level

45

Since this early work using Pyrex mortar bars – numerous researchers have confirmed the efficacy of lithium in controlling ASR in concrete with natural

• aggregates. One such study was a comprehensive research program conducted in the

UK using a large number of aggregates and concrete mixes with various SCMs stored under laboratory and field conditions

46

Most of the data generally follow the trends shown earlier and an example of this is shown in this graph. A lithium to alkali molar ratio in the region of zero-point seven to eight seems sufficient for lithium nitrate – whilst a slightly increased lithium dose is required for lithium hydroxide. This is because lithium hydroxide affects a small increase to the pH of the concrete pore solution and presents an extra challenge.

47

Recent studies confirm that lithium nitrate is the most efficient form of lithium for controlling ASR. As such a number of admixture suppliers currently distribute a lithium nitrate as a solution – in all cases the concentration of the solution is 30% lithium nitrate by mass. To achieve a lithium to alkali molar ratio of 0.74 requires just over half-a-gallon of 30-% lithium nitrate solution for each pound of equivalent sodium oxide in the concrete mix. This is approximately equivalent to 4.6 lires of solution for each kg of i l t d equivalent soda The amount of lithium may be reduced when it is used in combination with other preventive measures such as Class F fly ash – in such cases, advice should be sought from the admixture supplier

48

A fairly recent study conducted in Canada showed that the standard dose does not work with all reactive aggregates and that the efficacy of lithium was aggregate dependant. This graph shows data for 6 of 12 reactive aggregates studies. Lithium works well with these aggregates and the standard does seems to be sufficient – NOTE that 100% standard dose means a lithium-to-alkali ratio of zero-point-seven-four The 7th aggregate shown on the graph is a non-reactive limestone which doesn’t expand even when there is no lithium in the mix.

49

Unfortunately 3 of the 12 aggregates required more than the standard dose – 50% more in some cases …

50

Even this 50% increase was NOT sufficient with some aggregates and putting all the data together shows that the 100% dose gives extremely variable performance and is strongly dependent on the aggregate. Consequently … it is not possible to prescribe a fixed dose of lithium that will work with all aggregates and the amount

• f lithium required must be determined by testing with the specific aggregate being

used.

51

[no note required - this just reinforces in writing the narration for the previous slide]

52

In summary [presenter to read slide]

53